[RTTY] sound card

[Kok Chen]

On Jun 11, 2014, at 11:39 AM, peter.jackson17 at ntlworld.com wrote:

> I am going have a new computer built and I would like to know what  the best criteria for the sound card is please naturally with data modes in mind.

*How* you use a sound card is arguably more important than *what* sound card you use.

There is no point using a 24-bit sound card and clip the output.  You might as well use the cheapest sound card you can pick up at eBay if you are going to do that.   At the other end, if you let the signal fall below the noise floor of the sound card, you will not be able to decode either.

A lot of it also has to do with convenience.  If you don't mind manually riding the RF gain control of the receiver , pretty much any sound card will work.  If not, you will want a sound card that has at least 10 dB more dynamic range than the range of audio levels from the sound card.

The decoding error in digital modes, no matter whether it OOK CW or a 256 tone OFDM, is dependent on Signal-to-Noise ratio that the detector encounters, not on how strong the signal is. 

A sound card always generate its own noise.  Either from the analog circuitry before the A/D converter (ADC, codec, it comes in many names), or from the fact that it can only represent input voltages as a discrete number of voltages (this is called quantization noise).

Many 16-bit sound cards (e.g., the sound cards in MicroHAM interfaces) are noise limited by the codec noise floor.  24-bit devices typically have analog noise floors that are above the quantization noise because the quantization noise is so very low.

In general, you will want the noise floor of the sound card to be *at least 10 dB below* the sky noise (the hiss you hear when you connect an antenna to the receiver).  The reason I emphasize this margin is that when the sound card noise floor is equal to the sky noise, the noise that the FSK detector encounters is 3 dB worse.  The sound card noise therefore needs to be well below sky noise to reflect the true SNR of the signal.

So, whatever number you measure as the dynamic range from the receiver, you have to add 10 dB to get the required dynamic range from the sound card if you do not wish to ever have to touch the receiver's RF gain control.

With good software demodulators, you must *never* allow the sound card to ever clip.  If the sound card clips, the filters in the demodulators no longer work as they are supposed to.  You might as well use a lousy demodulator.

If you run with the receiver AGC turned on, the dynamic range requirement is much smaller (but it depends on whether the AGC works on the same passband as the signal that is sent to the sound card).  

Just put a DVM on your receiver's line output, with the bandwidth set to how you want to operate.  Tune to a quiet spot on the band, measure the RMS AC voltage, then find the loudest RTTY signal that you find, and measure the RMS AC voltage again.  The difference is the dynamic range from your receiver's line output.  

Add 10 dB to that, and you now know how much you need to spend on a sound card.

On top of noise, you can have other distortions that limits the useful dynamic range of a sound card.  The important ones are 3rd order IMD, and if you use wide filters and tune using a waterfall or a panadapter, the second order and third order hormonic distortion.

Below are numbers for some sound cards that I have measured, with the main tone at 1100 Hz, and second tone at 1200 Hz, and at 48,000 samples/sec.

Note that the microKeyer and Signalink both have more functionality than the sound card function.  

I know only of 2 hams who use the Presonus.  I know quite a number who use the microKeyer and SignaLink, and no one really admits to using a Syba :-)  (The Syba uses a C-Media CM119 codec).  

I have made an annotation in my notes that for my tests, the Syba was not stable if the source impedance is over 100 ohms.  For the Syba, I measured with  a 91 ohm generator impedance.  (Many transcievers' output impedance are much higher, and you must do something if you plan on using that inexpensive sound card.)  And because of the poor second harmonic distortion, I don't recommend it for PSK31 modems that tune using a waterfall.

Price:
-----------------------------------------
Presonus FireStudio Mobile   $300
MicroHAM microKeyer II    $500
Tigertronics SignaLink USB    $110
Syba UAUD    $8

Noise Floor (relative to full scale output):
------------------------------------------------
Presonus FireStudio Mobile    -117 dB
MicroHAM microKeyer II    -96 dB
Tigertronics SignaLink USB    -98 dB
Syba UAUD    -96 dB

3rd order IMD (measured at -6 dB full scale):
------------------------------------------------------
Presonus FireStudio Mobile    -87 dBc
MicroHAM microKeyer II    -65 dBc
Tigertronics SignaLink USB    -60 dBc
Syba UAUD    -88 dBc

2nd Harmonic distortion (measured at -6 dB full scale):
-----------------------------------------------------------------
Presonus FireStudio Mobile    -107 dBc
MicroHAM microKeyer II    -100 dBc
Tigertronics SignaLink USB  -98 dBc
Syba UAUD -81 dBc

3rd Harmonic distortion (measured at -6 dB full scale):
-----------------------------------------------------------------
Presonus FireStudio Mobile    -93 dBc
MicroHAM microKeyer II    -77 dBc
Tigertronics SignaLink USB  -64 dBc
Syba UAUD -105 dBc


73
Chen, W7AY